Robert J. Porte MD, PhD, Department of Surgery, Section Hepatobiliary Surgery and Liver Transplantation, University Medical Center Groningen, University of Groningen, P.O. Box 30.001, 9700 RB Groningen, The Netherlands
Traditionally, perioperative bleeding complications were a major concern during orthotopic liver transplantation, but a tremendous decline in transfusion requirements has been reported over the last decade. In recent years, there has been an increasing awareness towards perioperative thrombotic complications, including liver vessel thrombosis, and systemic venous and arterial thromboembolic events. Whereas a number of these thrombotic complications were previously categorized as surgical complications, increasing clinical and laboratory evidence suggest a role for the haemostatic system in thrombotic complications occurring during and after transplantation. High levels of the platelet adhesive protein von Willebrand factor with low levels of its regulator ADAMTS13, an increased potential to generate thrombin, and temporary hypofibrinolysis are all indicative of increased haemostatic potential after transplantation. Clinical evidence for a role of the haemostatic system in post-operative thromboses includes a higher thrombotic risk in patients with various acquired thrombotic risk factors. Although data on efficacy of anticoagulant therapy after liver transplantation are scarce, one study has shown a significant decrease in the risk for late hepatic artery thrombosis by antithrombotic therapy with aspirin. These findings suggest that antihaemostatic therapy in prevention or treatment of thromboembolic complications after liver transplantation may be relevant. Studies on efficacy and safety of these interventions are required as many of the thrombotic complications have a pronounced negative impact on graft and patient survival.
Since the first orthotopic liver transplantation (OLT) performed in the 1960s, improvements in surgical and anesthesiological techniques, pre- and post-operative management, and an increase in the donor pool have made OLT to a routine surgical procedure. However, historical bleeding was one of the major challenges during the procedure , nowadays more and more centres report transfusion-free liver transplant surgery . Massive bleeding complications during OLT still occur in some cases and remain a topic of concern. However, there is also increasing awareness towards perioperative thrombotic complications, which in cases may have a fatal outcome.
Traditionally, the liver transplant recipient was considered to be in a hypocoagulable state, as evidenced by abnormalities in routine haemostatic tests such as the platelet count, prothrombin time (PT), and activated partial thromboplastin time (APTT), and by massive perioperative transfusion requirements. However, the concept of the liver transplant recipient being hypocoagulable does not correspond to the current clinical observations of transfusion-free liver transplant procedures and an elevated risk of various thrombotic complications. Furthermore, the concept of ‘rebalanced haemostasis’ in patients with cirrhotic liver disease, which assumes that the ‘average’ patient with cirrhosis is in a new, yet relatively unstable, haemostatic balance because of a concomitant decrease in both pro- and antihaemostatic pathways, is increasingly appreciated [3, 4]. Although previously some of the thrombotic complications of liver transplant surgery have been attributed to surgical causes only, it may be that uncontrolled activation of the haemostatic system also contributes to the thrombotic risk, although this has been scarcely studied.
In this review we will summarize clinical and laboratory findings that, in our opinion, support the hypothesis of a hypercoagulable status in the liver transplant recipient which may contribute to perioperative and, early- and long-term, postoperative thrombotic complications. Furthermore, possible implications of these new insights for anticoagulant management will be discussed.
Thrombotic complications after OLT
Thrombotic complications occurring after OLT can be divided into local hepatic vessel thrombosis and systemic thrombotic complications.
Hepatic vessel thrombosis occurring in approximately 7% of patients undergoing OLT , pose a threat to both patient and graft survival. The incidence of hepatic artery thrombosis (HAT) is approximately 5%, although the risk in paediatric patients is much higher [5, 6]. Portal vein thrombosis (PVT) complicates around 2% of OLTs. HAT may occur early (within 2–3 months) after transplantation, but may also occur years after the procedure. Early HAT may result in necrosis of the bile ducts and eventually graft loss if arterial flow is not restored in time. In comparison with early HAT, late HAT might not be life threatening or even have clinical consequences because of the formation of collateral arterial circulation prior to the total obstruction. HAT is traditionally believed to be primarily of surgical origin. Indeed, technical imperfections at the anastomotic site, kinking of the artery, prolonged clamping of the hepatic artery, or the use of an arterial conduit increase the risk for HAT [7, 8]. Nevertheless, there is increasing data suggesting that changes in the haemostatic system may contribute to the development of HAT as well, as will be outlined below.
The risk of PVT after OLT is related to technical difficulties during surgery, prior PVT, a paediatric recipient, splenectomy, the use of venous conduits, and small portal vein size [5, 9]. Unlike HAT, which rarely occurs prior to transplantation, PVT is frequently seen in cirrhotic patients, with the risk of PVT increasing with the severity of the disease. Early PVT occurs in 1–2% of transplanted patients and causes acute clinical deterioration because of ischaemia, ascites and increased portal vein pressure . Early PVT is associated with an increased mortality compared with OLT recipients who do not develop a PVT [5, 11]. The risk of early PVT is increased (~3%) in patients who already had a PVT prior to OLT . In case of a late PVT, which occurs in ~1% of patients or in patients with a more slowly advancing stenosis of the portal vein, the symptoms can be mild.
Systemic thrombotic complications may occur in the perioperative period, but also years after transplantation. Intraoperatively, acute intracardiac thrombosis or pulmonary embolism (PE) may occur, with an estimated incidence of around 1–2% [12-14]. These complications are potentially fatal and appear to be more frequent in liver transplant recipients than in other surgical patients . In the immediate post-operative period, vena cava thrombosis occurs in approximately 0.4% of the recipients, and deep venous thrombosis in ~3% of patients .
Over time, the survival of liver transplant recipients has greatly improved with a current 5-year survival rate up to 82% depending on the indication [15, 16]. As an increasing proportion of patients have excellent long-term survival, long-term complications are becoming increasingly important. Besides immunosuppression-related malignancies, vascular complications significantly contribute to morbidity and mortality in the liver transplant recipient. Vascular diseases account for 21–50% of the mortality seen in patients surviving more than 5 years after transplantation, with the incidence of death as a result of vascular disease increasing over time [15, 17-19]. Borg and colleagues followed 311 patients surviving the first year after OLT . Of the patients that died during follow-up, 21% died as a direct consequence of vascular disease. The most frequent vascular diseases were myocardial (infarction, angina, sudden death, heart failure) or cerebral (transient ischaemic attack and ischaemic or haemorrhagic events). In a prospective study of OLT-recipients surviving the first year after transplantation, Abbasoglu and colleagues found that 45% of the patients that died during follow-up died from cardiovascular disease or cerebral vascular accidents . Rabkin and colleagues retrospectively reviewed OLT-recipients surviving at least 1 year after OLT . They found that 17% of the long-term mortality was because of cardiovascular disease (myocardial infarction and aortic aneurysm rupture). A recent meta-analysis of observational studies showed a 10-year incidence of cardiovascular events of 14%, which was 64% higher as compared with the general population. In particular, patients transplanted for non-alcoholic steatohepatitis appeared to be at risk for cardiovascular events post-OLT . This finding has been replicated in a recent single centre study .
Vascular events, including myocardial infarction, acute coronary syndrome, cerebrovascular accidents, and peripheral vascular occlusion, occur in approximately 7% of patients and fatal vascular events in around 3.5% of patients surviving the first year post-transplantation [15, 18].
Laboratory evidence for hypercoagulability during and after OLT
In recent years, it has become well established that the haemostatic system in patients with cirrhosis is in a rebalanced status as a result of a concomitant reduction in pro- and antihaemostatic systems [3, 22-24]. However, the haemostatic balance in patients with cirrhosis is much more fragile compared with individuals with an intact haemostatic system. Laboratory studies have shown that additional changes in the haemostatic system occur during OLT, the nature and relevance of which will be discussed below. After OLT it is presumed that the haemostatic system normalizes as liver synthetic function is restored, reflected by normalization of PT/INR values. Whether complete normalization occurs or whether clinically relevant differences compared with healthy individuals remain (for example because of immunosuppressive use) has been poorly studied.
In patients with cirrhosis the primary haemostatic system is frequently altered by thrombocytopenia. In addition, platelet function defects have been described , but modern platelet function tests, in which platelet function is tested in flowing blood, fail to identify these functional defects [26, 27]. The thrombocytopenia of cirrhosis is balanced in part by highly elevated levels of the platelet adhesive protein von Willebrand factor (VWF), although the functional capacity of this protein is somewhat compromised .
During surgery, haemodilution and platelet consumption as a result of haemostatic processes can cause a further decrease in platelet count. Platelet counts may also remain relatively stable even in the absence of platelet transfusion . During and up to 10 days after OLT, VWF levels remain substantially elevated . Interestingly, the functional capacity of VWF increases during the procedure, which is probably the result of release of VWF from endothelial cells as a result of activation of these cells by inflammatory processes and surgical stress. In addition, plasma levels of the VWF cleaving protease ADAMTS13 decrease during transplantation, resulting in a profound dysbalance in the VWF/ADAMTS13 ratio, which has been linked to thrombotic risk in different disease states such as sepsis and myocardial infarction [31, 32]. Using a platelet function test in flowing blood, we demonstrated an enhanced capacity of plasma from patients taken during and after OLT to support platelet adhesion, which is likely linked to this profound VWF/ADAMTS13 imbalance . In addition, we have provided evidence that the platelet itself remains functional during and after OLT . Given the thrombotic phenotype associated with isolated ADAMTS13 deficiency or elevated levels of VWF, we hypothesize that the preserved platelet function, in combination with the prothrombotic VWF/ADAMTS13 imbalance results in a normal or even supranormal functional status of the primary haemostatic system. We hypothesize that the normal or hyper-reactive primary haemostatic status may constitute a risk for post-operative thrombotic events, but this has never been directly clinically examined. However, indirect clinical evidence for a role of activation of primary haemostasis in post-transplant thrombosis comes from a study in which platelet inhibition by aspirin was shown to substantially reduce the risk of HAT (see below). Interestingly, in our study on VWF and ADAMTS13 during and after OLT, the patient who had very low levels of ADAMTS13 during and after surgery developed an early HAT . Future, adequately powered studies should examine whether a post-operative VWF/ADAMTS13 dysbalance indeed is a risk factor for early HAT. Although the risk of post-operative HAT is likely initiated by local activation of endothelial cells as a result of the ischaemia-reperfusion injury inevitably linked to a transplant procedure  and by other surgical risk factors involving the arterial anastomosis [7, 8], we believe that a hyper-reactive primary haemostatic system may enhance this risk even further. The local endothelial activation and damage in combination with a hyper-reactive haemostatic system will result in platelet adhesion, which may progress to an occlusive hepatic artery thrombus.
Secondary haemostasis in patients with cirrhosis is characterized by a commensurate decline in both pro- and anticoagulant proteins. Traditionally, the net effect of the haemostatic changes in secondary haemostasis in cirrhosis was believed to be a hypocoagulable status as evidenced by prolongations in routine diagnostic tests of coagulation such as the PT and APTT. Although the prolonged PT and APTT suggest a bleeding tendency, these tests fail to take into account the reduced levels of the anticoagulant proteins. When examined with more sophisticated tests, coagulation potential, as tested by thrombin generation assays, is indistinguishable from or even superior to that of healthy controls [35-37]. The normal or enhanced thrombin generation in cirrhosis has been attributed specifically to increased plasma levels of coagulation factor VIII with decreased plasma levels of protein C [36, 38]. A normal to hypercoagulable status in patients with cirrhosis has also been demonstrated using thrombo elastography .
During OLT, levels of pro- and anticoagulants decrease even further, which is associated with a further prolongation of PT and APTT. Nevertheless, the coagulation potential as tested by thrombin generation assays remains comparable to or is even increased as compared with that of healthy volunteers . Importantly, in the post-operative period (at 5 and 10 days after surgery), in vitro thrombin generation tests showed an increased rate and total amount of thrombin generation. Decreased levels of protein C and S and persistently high levels of FVIII are likely related to this hypercoagulable status. Whether this in vitro hypercoagulable status has a direct link to thrombotic complications has not yet been studied and requires further research.
The fibrinolytic system in patients with cirrhosis is also in a rebalanced status, but during surgery a hyperfibrinolytic status may occur temporarily as a result of an increase in plasma levels of tissue-type plasminogen activator (tPA) in the anhepatic phase and after reperfusion . However, at the end of surgery, a hypofibrinolytic status occurs because of a massive increase in plasminogen activator inhibitor type 1 (PAI-1) levels, and the fibrinolytic potential remains significantly decreased up to day 5 after surgery . The link between this temporary hypofibrinolytic status and the risk of thrombotic complications has not yet been studied. Nevertheless, a hypofibrinolytic status has been demonstrated to be a risk factor for venous and arterial thrombosis in the general population .
Altogether, these laboratory data are suggestive of a rebalanced haemostatic system or even hypercoagulability, in particular in the early post-operative period. It must be pointed out, however, that the limited data available concerns patients transplanted for cirrhosis. To our knowledge, data on haemostatic changes during and after OLT in non-cirrhotic patients (e.g. patients with hepatocellular carcinoma without cirrhosis, or patients with metabolic liver diseases) are lacking in the current literature, but as 80% of the liver transplants performed in the Western world are performed in cirrhotic patients, we feel this limitation is acceptable.
Thrombotic complications during and after OLT – Is there a role for the haemostatic system?
The various thrombotic complications that may occur in the perioperative period have long been assumed to be of surgical origin or related to graft quality. In addition, a role for the haemostatic system in early thrombotic complications has long been dismissed because of the traditional belief that a hypocoagulable status is present prior to and during OLT. Thrombotic complications that occur long after transplantation have been linked to graft or environmental factors (e.g. late HAT), or are believed to be secondary to immunosuppression (e.g. systemic arterial events) . Nevertheless, there is accumulating clinical evidence for a role of the haemostatic system in both early and late thrombotic events. Most data concern the role of the haemostatic system in the development of HAT, but the hypercoagulable status may also (in part) be involved in the development of PVT.
First of all, patients transplanted for metabolic liver diseases such as familial amyloidotic polyneuropathy (FAP) or acute intermittent porphyria (AIP), appear to have an increased risk for early HAT compared with patients transplanted with fibrosis or cirrhosis [44, 45]. In contrast to patients with cirrhosis, patients with FAP or AIP do not have a compromised liver function, and therefore have an intact haemostatic system. Moreover, the procedure in these patients is technically easier as compared with the procedure in patients with cirrhosis in whom portal hypertension, venous collaterals or perihepatic inflammatory adhesions complicate the hepatectomy. Thus, despite a less complex surgical procedure, patients with FAP or AIP have an increased risk for HAT, for reasons that have not been clearly defined. We hypothesize that the increased risk for HAT in these patients is directly linked to their intact haemostatic system. Although the surgical triggers for development of HAT are decreased in patients with FAP or AIP, the haemostatic capacity is increased compared with that of patients with cirrhosis, thus explaining the increased thrombotic risk.
Secondly, patients receiving a graft that carries a thrombophilic mutation in one of the coagulation factors (such as Factor V Leiden or the prothrombin G20210A mutation) appear to have an increased risk for HAT [46, 47]. As biochemically and clinically these mutations clearly increase coagulation potential in the general population, we believe that these findings indicate an important role for the coagulation system in development of HAT. The observation that the risk of HAT is increased by genetic factors that increase hemostatic potential in patients transplanted for cirrhosis is compatible with the observation that the risk of HAT is increased in patients transplanted for metabolic diseases, who have a more competent haemostatic status as compared to a patient with cirrhosis.
Thirdly, it has been demonstrated that cytomegalovirus (CMV) infection is associated with an increased risk of HAT . Multiple studies have shown the risk of HAT to be substantially increased in seronegative recipients receiving a seropositive graft [8, 48-50]. We hypothesize that this finding is, at least in part, related to the prothrombotic effects of CMV infection. CMV infection results in endothelial cell activation which in turn results in increased platelet reactivity and procoagulant capacity of these cells . Notably, some other studies have failed to show the correlation between HAT and CMV infection. One study investigating risk factors for early HAT did not find CMV infection in donor or recipient to be associated with early HAT after OLT . Another study found a correlation of CMV with late HAT, but not with early HAT . It must be pointed out that in both these studies patients were intensively monitored and treated for CMV viraemia. The early treatment might delay or stop the endothelial damage and activation of the haemostatic system as a result of the CMV infection.
Forth, it has been demonstrated that patients with PVT before transplantation have an increased risk for development of HAT after transplantation . This may indicate a certain pre-existent hypercoagulable status unrelated to the liver, as the hypercoagulable factor persists after transplantation and causes an increased risk for HAT [30, 40]. On the other hand, endothelial damage of the portal vein wall will persist even after successful thrombectomy during transplantation, and this could contribute to the higher risk of recurrent PVT in these patients as well.
Fifth, it has been demonstrated that indefinite treatment with aspirin results in a >80% relative risk reduction in late HAT in ‘high risk’ patients, although this effect has only been reported in a single, retrospective study . These results do not necessarily imply that HAT is in part caused by a hypercoagulable status, however, they do indicate that the haemostatic system is somehow involved in late HAT. In addition, many other factors such as donor age, severe acute rejection, backtable surgery for anatomic variations, blood group-incompatible grafts, active cigarette smoking, usage of a donor iliac artery interposition graft to the aorta, and use of a graft from a donor who died of a cerebrovascular accident may have a role in the development of HAT , but none of these factors are modifiable, whereas the administration of platelet inhibitors is a straightforward way to inhibit the primary haemostatic system.
Altogether, in our interpretation of the clinical literature there is a clear role for the haemostatic system in development of both early and late HAT. Moreover, although the data are limited, there is laboratory evidence for a hypercoagulable status in the perioperative period. This hypercoagulable status may not only contribute to the risk of HAT or PVT, but may also contribute to other post-operative thrombotic events including deep vein thrombosis. Although the incidence of venous thrombosis after OLT is not necessarily higher than in other types of major surgery, the fact that these complications do occur despite hypocoagulable routine laboratory tests (platelet count, PT, APTT) in the first post-operative days indicates that these routine tests probably do not reflect in vivo physiology. On the other hand, part of the venous thrombotic events that occur after OLT may be a consequence of insufficient or absent thromboprophylaxis, which is withheld because of a presumed bleeding risk.
Liver transplant recipients have an increased risk for cardiovascular disease and risk of death because of cardiovascular events years after the transplant . The increased risk for cardiovascular events is likely because of metabolic changes which are presumably mainly related to immunosuppressant use. In particular, diabetes mellitus, hypertension, dyslipidemia, and obesity occur frequently in the transplanted patient [55, 56]. In addition, the risk appears higher in patients with pre-existing cardiovascular risk factors, such as in patients transplanted for NASH [19, 21]. Nevertheless, it cannot be excluded that activation of haemostasis by a dysbalanced VWF/ADAMTS13 axis, a hypercoagulable or a hypofibrinolytic status, which are established risk factors for arterial events in the general population, also contribute to the elevated risk of arterial events in the transplanted patient. Notably, metabolic risk factors themself already result in procoagulant changes. For example, patients with hypertension or the metabolic syndrome outside the context of liver transplantation have significant changes in their haemostatic system [57, 58]. It will thus be difficult to disentangle coagulation disturbances because of immunosuppression-related metabolic alterations from isolated coagulation abnormalities. Evidence for direct activation of prohaemostatic components by immunosuppression was recently summarized by Senzolo and colleagues . Several studies regarding immune suppressive medication and haemostatic changes were discussed, among which a study by Huang and colleagues who demonstrated increased levels of VWF and PAI-1 after addition of dexamethasone to endothelial cells . Another study, performed by Bombeli and colleagues demonstrates an increase in thrombin generation by endothelial cells exposed to cyclosporine .
Regardless of the origin of the prothrombotic status, the fact that hypercoagulability likely contributes to the occurrence of various thrombotic events after transplantation means that studies on the prevention of such events by means of prophylactic antithrombotic therapy are indicated.
A timeline with the various thrombotic complications that may occur prior to, during, and after OLT is depicted in Fig. 1. The figure also depicts the haemostatic status of the liver transplant recipient over time.
Possible prophylactic antithrombotic strategies to reduce the incidence of post-transplant thrombotic complications
In the previous paragraphs we have summarized the increasing clinical and laboratory data indicating a role of the haemostatic system in both early and late thrombotic complications after OLT. Although perioperative bleeding complications occur more frequently as compared to post-operative thromboses, the bleeding complications are mostly manageable, whereas post-operative thromboses (such as hepatic vessel occlusions) are associated with substantial morbidity and even with mortality. Therefore, it is our belief that prophylactic antithrombotic treatment to prevent these complications may be clinically relevant, in particular as other risk factors related to thrombotic risk (e.g. those related to graft quality or immunosuppression) are not easily modifiable. Traditionally, antithrombotic treatment is cautiously used in liver transplant recipients because of the perceived bleeding risk, in particular in the immediate post-operative period. We propose that the recent novel insights in the haemostatic status after OLT should lead to a cautious reconsideration of the use of antihaemostatics. Various antithrombotic strategies may be considered, and each of the potential regimens need a risk/benefit assessment, preferably in controlled clinical studies. All antithrombotic strategies are associated with a risk for bleeding complications, which will require careful attention in clinical studies. Below we will discuss possible antithrombotic regimens and their potential indications. We would like to stress that these suggestions are not meant as clinical guidelines, but rather serve as considerations for future, carefully controlled, clinical studies. None of the proposed regimens have been clinically validated, but we believe the time has come to start considering clinical validation of some of these options.
Standard low-molecular weight heparin (LMWH) thromboprophylaxis might be helpful to avoid venous thrombosis, but the partial antithrombin deficiency in the early post-operative period is of concern, as the anticoagulant activity of heparins relies on antithrombin. Furthermore, it has recently been demonstrated that LMWH has a stronger anticoagulant activity in plasma from patients with cirrhosis as compared with plasma from healthy individuals . In addition, monitoring of LMWH is complicated as the antiXa laboratory test appears to overestimate the true LMWH mass . The new generation direct Factor Xa or thrombin inhibitors have advantages as they do not rely on antithrombin, have no risk for heparin-induced thrombocytopenia, and can be continued for a prolonged period. However, a disadvantage may be that these drugs are metabolized in the liver and kidneys, which will result in unpredictable anticoagulation in patients with delayed graft function or post- operative renal failure. LMWH thromboprophylaxis may also help to prevent early PVT or HAT, but to our knowledge the effect of standard thromboprophylaxis on PVT or HAT has never been assessed in the post- transplant population. It has been recently demonstrated that LMWH anticoagulation prevents PVT and decompensation in patients with cirrhosis prior to OLT . Notably, LMWH does appear to have a favourable safety profile in patients with cirrhosis . LMWH may be switched to oral anticoagulation with vitamin K antagonists with the aim of avoiding hepatic vessel thrombosis weeks or months after the transplantation. Drawbacks of oral anticoagulation in the liver transplant recipient are possible interactions with immunosuppressive drugs.
In a single study, long term aspirin has been shown to lead to a reduction in the incidence of late HAT in selected patients, without inducing bleeding complications. The risk/benefit of this approach should be prospectively investigated in a more general cohort of transplanted patients. Long-term aspirin may also lead to a reduction in cardiovascular events, which may be particularly relevant for patients transplanted for NASH, but admittedly studies to assess this will have to be large and lengthy.
Short-term anticoagulation using combination therapy may be of additional benefit to prevent early events, in particular PVT or HAT. A combination of LMWH and aspirin might be relevant and relatively safe. More intensive antihaemostatic regiments, including dual or triple antiplatelet therapy with aspirin, P2Y12 blockers (such as clopidogrel) and/or αIIbβ3 blockers (such as abciximab) may be relevant to avoid (arterial) events in the early post-operative period, but the bleeding risk may be significant.
The choice for a moderate or more intensive anticoagulant regiment may determine on the patients' individual risk profile. For example, in patients with pretransplant PVT, thrombotic complications prior to transplantation, a paediatric recipient, or a pre-existing cardiovascular risk profile one might consider a more intensive regimen. In patients with severe bleeding during OLT, massive hyperfibrinolysis and no thrombotic history, a more cautious approach may be warranted. Also, the duration of anticoagulation may depend on patient characteristics. The patients with a very unfavourable cardiovascular risk profile may benefit from a more intensive treatment as compared to the patients without diabetes, obesity, and with relatively well-controlled hypertension and lipid profile.
Here we have summarized the increasing laboratory and clinical data indicating a role for the haemostatic system in the development of thrombotic complications during and after OLT. Clinical observations also suggest a hypercoagulable haemostatic system in the long term after OLT, but additional laboratory studies are required to assess the haemostatic status in the transplanted patient in detail. We believe that anticoagulant strategies to prevent the various thrombotic complications require serious consideration, but the risk/benefit profile of the various strategies in the complex post-transplant patients is still unclear. Large and long-term clinical studies to assess the efficacy of the various antithrombotic strategies to avoid the in cases fatal thrombotic episodes will be required. Although antithrombotic regimens inevitably carry the risk of bleeding complications, these may be justified by the severity of some of the thrombotic syndromes.